NTN TECHNICAL REVIEW No.

732005

[ Technical Article ]

Application of Topology Optimization and Shape Optimization for Development of Hub-Bearing Lightening

Haruo NAGATANI* Tsuyoshi NIWA*

From the perspective of global environmental protection and conservation of resources, the development of lighter automobile parts for fuel cost savings is a worldwide initiative. In this abstract, we seek the ultra-lightweight Hub-Bearing, and succeed in its development. In the development of this product, two optimization methods are used: First is a topology optimization; the second is shape optimization. We introduce the process of the development by explaining the two optimization methods shown above.

1. Introduction
The enactment of the US CAFE (Corporate Average Fuel Economy) regulations in 1975 triggered increased demand for lightweight designs for every automotive component. Lighter components lead to better automobile fuel efficiency, which in turn contributes to energy conservation and global environmental preservation. In manufacturing industries, various efforts to achieve lightweight designs are in progress and bearing manufacturers are committed to decreasing the weights of hub bearings.1) 2) NTN has also been working to lighten various products. By applying shape optimization techniques to the development of lightweight 3rd generation hub bearings, we have succeeded in dramatic weight reductions. This report describes our optimization techniques as well as examples of lightening effects.

the lightest in the world. In the initial stage of development, a prototype hub bearing weighing 1.3 kg failed a durability test, while one weighing 1.4 kg, illustrated in Fig. 1, succeeded in achieving functional targets including for mechanical strength. This result marked the starting point of our analysis for optimal bearing shape.

Outer ring Hub shaft Inner ring

2. Development goal
The hub bearing being developed is for low weight cars and its target mass is 1.0 kg, which would make it
*Automotive Sales Headquarters Automotive Engineering Dept. Fig.1 Hub-Bearing original design

-14-

Application of Topology Optimization and Shape Optimization for Development of Hub-Bearing Lightening

3. Overview of shape optimization

The basic concept of shape optimization design is to place material in areas that truly need it and thin out unnecessary material from areas that are not important for correct function in order to obtain the minimum shape that satisfies all the necessary functional requirements, such as mechanical strength and rigidity. Spurred by the lightening needs mentioned above, recently the demand for the ability to determine optimal shapes easily has been mounting. However, currently available software packages are not powerful enough to fulfill this demand. The currently available analysis techniques can be roughly categorized into (1) topology optimization and (2) shape optimization (these techniques will be described in detail later). For topology optimization, an optimal structure for carrying an external load is determined, while for shape optimization, the areas to be altered and the ranges for possible dimensions are set in advance and optimal values are determined accordingly. Since each type of technique has advantages and disadvantages, we should utilize the strengths of both to attain our goals. In our development project, topology optimization was executed first to develop a rough shape (basic shape), and then adjustments were made as necessary to achieve optimal dimensions that do not greatly differ from those of that basic shape. The reasons for this two-step scheme are as follows.

between a flange and a brake rotor or between an inner ring and a hub ring are treated. For this reason, high-precision nonlinear analysis is necessary to verify the results obtained from topology optimization. (2) Optimal shapes obtained from topology optimization are complicated 3D forms that may not be obtainable by machining at reasonable costs or may not be obtainable at all. Therefore, a process to develop a machining-ready shape based on the results of topology optimization is necessary. This new basic shape is again subjected to shape optimization to realize the final optimal shape.
*Certain software packages are now capable of analysis using gap elements. However, even with these packages, setup for an object with a complicated shape is time-consuming, and areas with complicated profiles must be set as unalterable. As a result, these packages are insufficient in terms of analytical accuracy.

4. Topology optimization (1st step)

In the topology optimization step, an optimal structure to achieve the functions necessary for the part in question is obtained. In some cases, dramatic changes in the structure are possible. We used a homogenization method 3) as a topology optimization technique.

Analysis model
The analysis model we used is illustrated in Fig. 2. The hub ring was integrated with the inner ring, mesh was plotted and the 3 translational degrees of freedom for the bolthole nodes on the flange were constrained. For the external load working on the hub ring, load concentration was assumed to occur on the rolling

Necessity of two-step analysis scheme

(1) Topology optimization analysis is essentially linear analysis,* and does not offer sufficiently accurate results for a boundary nonlinear problem such as stress analysis for hub bearings where contacts

Hub ring

Flange: design region

Bolt-fastening section: non-design region

Inside diameter section: Outside non-design region Inside design region

Shaft: non-design region

Inner ring

Wheel spigot joint outside diameter: non-design region

Inner side view

Outer side view

Fig.2 Topology optimization model

-15-

NTN TECHNICAL REVIEW No.732005

surface and a rolling element load value obtained from a bearing internal force analysis program was utilized. As shown in Fig. 2, regions where the shape can be altered (design regions) were set larger, and regions where the shape cannot be changed (such as the wheel spigot joint outside diameter and bearing outside diameter) due to interactions with counterpart members were defined as unalterable regions (nondesign regions). Furthermore, analysis was executed considering the forging draft, cyclic symmetry conditions of boltholes and symmetricalness relative to the straight lines connecting the boltholes and the flange center. The object function and restricting conditions were as follows. Objective function: Distortion energy minimization Restricting conditions: 22.5%, 27.5% and 30% relative to the volume (or mass) of initial shape The analysis was performed with two hub bolt patterns - 0 (cross-pattern) and 45 (X-pattern) relative to the vertical direction. Since the stress was greater in the 0 direction, further analysis was limited to this direction.

Analysis results
The shapes and principal stresses shown in Figs. 3 and 4 were attained. As shown in Fig. 4, with greater weight reduction, stress rapidly increased. We adopted a shape that was 27.5% the initial volume and near the stress limit as the basis for the basic structure of the hub bearing.

Maximum stress MPa

0.1

0.2

0.3

0.4

Proportion of initial volume

Fig.4 Principal stress of each design

Indicates the initial shape. (See Fig. 2.)

Wheel spigot joint outside diameter (non-design region)

30% the initial volume

27.5% the initial volume

22.5% the initial volume

Fig.3 Result of topology optimization

-16-

Stress limit value

Application of Topology Optimization and Shape Optimization for Development of Hub-Bearing Lightening

5. Shape Optimization (2nd Step)

Analysis model
Using the shape obtained from the previously described topology analysis as a basis, the basic shape was set as in Fig. 5 and shape optimization analysis was executed by varying the dimensions at the seven locations (h, b, y1, x2, y2, x3, y3) shown in Fig. 6. The fully automatic analysis system summarized in Fig. 7 was developed and the analysis was executed based on the L27 experiment plan method. Objective function: Volume (mass) minimization Restricting conditions: Principle stress at three designated nodes (yellow dots in Fig. 6) below the current level The software of the shape optimization analysis system drives the nonlinear analysis solver to

calculate stress values while altering the shape according to the mesh data. In altering the shape, the shape base vectors* were set using the morphing software so that alteration of the mesh could be synchronized with alteration of the shape in order to maintain the cyclic symmetry of the bolt holes as well as the symmetricalness relative to the straight lines connecting the bolt holes and the flange center. The calculation was executed on a 2.8 GHz Pentium 4 computer and took 2.5 days to complete. From this result, a response curved surface approximation model was developed to achieve an optimal solution.
*Vectors that defines how each node shifts when the nodes used as parameters are shifted.

Nonlinear analysis software Morphing software Optimization software Shape base vector setup Design parameter setup
Objective function: analysis result
Restricting conditions Read operation Drive operation
Execution of nonlinear analysis

FEM model

Write operation

Analysis result

Optimal solution

Fig. 5 Basic design of shape optimization (As installed to the flange of test rig)

Fig. 7 Shape optimization systems

Height

Height: h Fixation at end point of tangential line

Width: b

Width: b y1<Y1 where Y1 is current value of y1

h,y1,b x2,y2,b x3,y3,b

h,y1,0 X4=const. h, x2, x3<X4 x2,y2,0 x3,y3,0 Z X x4,y4,0

Fixed

Fig. 6 Parameter of DOE

-17-

NTN TECHNICAL REVIEW No.732005

L27 best solution

Optimization

Initial value

L27 best solution

Optimization

Stress point 1 (a) Volume

Stress point 2 (b) Stress value

Stress point 3

Fig. 8 Result of shape optimization

Analysis results
In Fig. 8, (a) summarizes volumes, while (b) shows stress values. "L27 best solution" represents the combination that performed most effectively in the calculations performed according to the experimental method while "Optimization" represents the optimal solution obtained by calculation in accordance with the response curved surface approximation. As a result of this series of calculation operations, many of the optimal values obtained coincided with the upper or lower limits (limits in terms of avoidance of interaction with counterpart parts). Consequently, as can be understood from Fig. 8(a), there was no apparent difference between the L27 best solution and the optimal values obtained from the response curved surface. We cannot deny the possibility of obtaining better optimal values, but we believe that we have successfully determined values very near the ideal optimal values. Fig. 8(b) summarizes stress values associated with varying shapes. At every evaluation point, the stress value is lower than the current level. From these findings, we determined the shape of our final design.

achieved the target mass of 1.0 kg and cleared the target values for strength, durability and rigidity. In addition, the hub ring and outer ring have also achieved lightening targets as illustrated in Fig. 10.

Fig. 9 Shape of final design

Outer ring

6. Final Design
In the development work, we also applied the analysis technique described above to the outer ring. In addition to shape optimization, we attempted to achieve lightening in the finalized bearing specification through alteration of bearing internal design and development of new materials and grease. The newly developed hub bearing shape, which is illustrated in Fig. 9, has
100g

Hub ring

240g Fig. 10 Amount of each component Lightening

-18-

Application of Topology Optimization and Shape Optimization for Development of Hub-Bearing Lightening

7. Conclusion
This report has introduced the shape optimization analysis executed for lightening hub bearings for reduced weight cars. With this technique, we have succeeded in achieving a target mass of 1.0 kg, attaining dramatic lightening that was previously considered impossible. Note, however, that this newly developed hub bearing is not suitable for drum brake structures in which the hub bearing also functions as a brake seal.* For such applications, another hub bearing type developed through use of our shape optimization technique should be used (at a penalty of additional 50 g weight). Our analysis in this report was centered on mechanical strength. However, we also need to set targets for hub bearing rigidity. Therefore, we will aim for multi-faceted optimization in which we attempt to promote further lightening while maintaining sufficient bearing rigidity so that we can establish hub bearing technology to cope with the needs of various car manufacturers.
*As shown in Fig. 9, the newly developed hub bearing has four pawls that radially guide the members installed to the hub ring. If the hub bearing is applied to a drum brake, the pawls need to be replaced with a circumferential rim to provide sealing function.

References
1) J. Sakamoto, "Hub unit bearing market and technology trends," Monthly Tribology, October 2004 2) K. Kajihara, "Improved hub unit analysis simulation techniques," Koyo Engineering Journal No. 167 (2005)

Photos of authors

Haruo NAGATANI
Automotive Sales Headquarters Automotive Engineering Dept.

Tsuyoshi NIWA
Automotive Sales Headquarters Automotive Engineering Dept.

-19-